The cubicles were arranged in quads, four seats each, and there were three quads in the room, plus a tiny kitchen area and an office for Melody. That made twelve team members, including Melody, with me as the thirteenth. Not that I was superstitious or anything.
Melody took me on the tour, introducing me to each of the members of the team. They were mostly young, and friendly enough, but I forgot their names almost as soon as I met them. Names had never been my strength.
I did notice that seven of the twelve were women, a higher percentage than I had expected, and what I would later find out was well above the NSA average. I mentioned it to Melody, who shrugged and said, “Women don’t posture. There’s no room on this team for personal ambition or for trying to appear to be something you’re not. If I get even a hint of that, I’m not interested. Your average male intelligence agent, well…”
“I’ll keep that in mind,” I said.
“This will be your quad,” Melody said. The other three chairs were taken by an older white man with a fringe of white hair at the back of his head, an Asian woman with a slender face and large glasses, and Shaunessy Brennan.
“You’ve got to be kidding me,” Shaunessy said.
Melody ignored her. “And this will be your seat.” The fourth spot had a sagging swivel chair with a broken arm, and no computer.
“You mean you vouched for him? After he pulled that stunt?” Shaunessy said.
“We were told to hack into the account,” I said, for what seemed like the twentieth time. “I was following instructions.”
“He’s part of the team, for better or worse,” Melody said, and that seemed to end the discussion. To me, she added, “We’ll get you a machine and a better chair. Office furniture requisitions take forever around here. You’ll need your own account as well.”
“Don’t worry, he can just hack into somebody else’s,” Shaunessy said. “Though I don’t know why he needs a computer at all. Just give him a pencil and a stack of paper.”
“Why don’t you come with me for now, and we’ll get you set up as best we can,” Melody said.
I followed her back to her office. It was crowded with knickknacks, mostly geekware of some kind or another. I saw a binary clock, a chess set with a half-finished game, and a plush Cthulhu. Her bulletin board had a photo of a little girl—I was guessing a granddaughter—dressed as Chewbacca, and a hand-lettered sign that read TANSTAAFL.
Melody sat in her swivel chair with the same elegance she might have if it were a throne. “Welcome to the team,” she said formally.
I took the chair opposite her desk. “Thank you. Just what team is this?”
“The team of misfits,” she said. It seemed to be a joke, but she didn’t smile. “I’d like to say we do the jobs nobody else can do. But often we just do the jobs nobody else wants to do.”
I didn’t answer. My eyes roamed the office and settled on the chess set. White was a knight up, but its pawn structure had been demolished. I liked Black’s chances better.
“The vast majority of all traffic these days is encoded with public key encryption,” Melody said. “Which, I’m sorry to say, is unbreakable.”
I waited. The whirring noise in the background was the only sound. “As far as most people think,” I prompted her.
“I’m afraid not,” she said. “RSA encryption really is unbreakable, if it’s done right. We have more compute power than anyone else in the world, and we can’t touch it.”
“But this is the NSA,” I said.
She sighed. “Then I guess your disillusionment starts here. You’re a mathematician. Do I really have to give you a primer on big numbers?”
“But… the NSA,” I said.
“2048-bit encryption. The kind your phone can manage in a few milliseconds.
How many possible keys is that?”
“22048,” I said immediately.
“Which means that to find your key in a brute force attack, I need to make 22048 guesses. Or half of that, on average.”
“But you don’t brute force it,” I said. “Come on, you’ve got the Sieve of Atkin, at the very least, to narrow the guesswork, and you’ve probably got a lot better tricks than that.”
“We do,” she said, the hint of a smile playing around her lips. “But bear with me. Say I have a computer that costs one dollar and can make a billion guesses a second. We’re not even in the ballpark there, but let’s just imagine such a computer exists. How many computers would I need to guess your key in, say, a million years?”
“Brute force?” I did the math. “A billion is 230, give or take. A year is maybe 225, so a million years is 245… altogether, call it 21973,” I said.
“And how much do you think we can knock off with prime sieving?”
“That allows you to skip all the non-prime numbers,” I said. “The rule of thumb is an average separation of 2.3 primes per digit, so with numbers of that size, I’m going to guess you have a prime every, what, one or two thousand?”
“Which brings it down to what?”
“21969,” I said. “Ish.”
“And do you think we have 21969 computers?”
My faith in the NSA was waning, and I was starting to feel a little foolish. “No.”
“Or 21969 dollars to buy them?”
I sighed. “I get the picture. I just thought that there would be, you know, another way. That somebody would have invented something by now to crack it.”
Melody smiled beatifically. “Now don’t despair. I said properly encrypted messages were unbreakable, but, fortunately for us, very few messages are properly encrypted. Even now, less than ten percent of HTTP traffic goes through SSL, and of those that do, the vast majority use the primes hardcoded in their key exchange software. Software packages that do the encryption can have bugs, which we can exploit if we find them or know about them. Also, although the message might be sent encrypted, it has to be decrypted on the other side, and that computer system itself might be vulnerable to attack.”
“So… that’s what this team does?”
“No. The NSA already has thousands of hackers, algorithm experts, and mathematicians who work on those problems. We have experts on every operating system and software package out there, teams that invent ways to identify non-randomness, teams that look for ways to knock an order of magnitude or two off of the time it takes to crunch through a trillion keys.”
“But not us.”
“No. Nothing so banal. We work on messages that aren’t public key encrypted. That’s a very small percentage of the overall traffic. Usually, that means they use some old tried and true method, and can be cracked by a Raspberry Pi with one hand tied behind its back. Occasionally, though, we get messages that aren’t encrypted with any recognized variation of obsolete technique, and we can’t read them. I’ll be honest with you: most of those are never cracked. But we’re the team that gets to try.”
I couldn’t help grinning. “Sounds like just my kind of team.”
“It’s not glamorous, and it’s not all that valued. I have to constantly fight to keep funding. A lot of the messages we wrestle against probably aren’t even meaningful at all—just scrambled signals that were encoded wrong or garbled by atmosphere or bad equipment. But every once in a while, we crack something that nobody else could.”